Join 3,572 readers in helping fund MetaFilter (Hide)


Download vs. store-bought games. Which is hotter?
September 4, 2014 2:05 PM   Subscribe

Downloading video games from the Internet creates a larger carbon footprint than driving to the store to purchase the same game on a Blu-ray disc, according to findings published in the Journal of Industrial Ecology.
For an 8.8-gigabyte PlayStation video game file — the average size of video games in 2010 — the resources required to produce, distribute, and dispose of Blu-ray discs are far less than the energy required to power servers, routers, and networks involved in downloading the game file, researchers say. The advantages of discs decrease as file sizes shrink, the analysis found, and for game files less than 1.3 gigabytes, downloading has a smaller carbon footprint than purchasing the game on Blu-ray. (via)
Gameplay (use) still accounts for 93% of lifetime emissions, and as the authors observe, "The results of this study are unlikely to change or influence consumer behaviors; the carbon emissions of production and distribution are not known to be factors consumers consider when buying electronic goods or entertainment media." One might imagine online movie rentals like Netflix has a similar profile vs. in-store rentals.
posted by stbalbach (55 comments total) 19 users marked this as a favorite

 
I suspect "average" is a useless measure these days. It's bimodal: indie games that are often under a gig, and (for example) The Last of Us: Remastered for PS4 at a whopping 50 GB. Ok now I'll read the thing.
posted by 2bucksplus at 2:07 PM on September 4 [1 favorite]


It is right there in this post: "game files less than 1.3 gigabytes, downloading has a smaller carbon footprint than purchasing the game on Blu-ray"
posted by soelo at 2:13 PM on September 4 [3 favorites]


I skimmed but the point that caught my eye was the fact that it's assumed when buying the game at retail, you are also buying nine other items, so the carbon use of driving to the store is divided among those ten items. I rarely buy discs anywhere I am also induced to buy further merchadise, maybe big box grocery retailers, Gamestop/Best Buy, not so much.
posted by Keith Talent at 2:17 PM on September 4 [7 favorites]


Really? It seems like there's practically zero marginal carbon emission for data transmission since it's all pretty much fixed infrastructure. The Blu-ray disc is always marginal production - every new copy requires a new disc.

Is it all from electricity consumption? I mean, how much marginal electricity is consumed by a 2GB download, even all the way from server to user? It can't be more than a handful of watts.

On the other hand, if you're amortizing base utilization across the number of GB transmitted in a day, sure. But I'm not sure which approach is more correct.
posted by GuyZero at 2:25 PM on September 4 [2 favorites]


If we make a really optimistic set of assumptions on one side, and then make the most pessimistic set of choices on the other, the results will not be what you expect. Click here to give us page views!
posted by Nerd of the North at 2:25 PM on September 4 [69 favorites]


Weber and colleagues (2010) estimate that carbon equivalent emissions of compact disc (CD) production and distribution in the United States is almost 3 times8 that of PS3 BD production in Europe. The higher emissions for optical disc production in the United States are mainly the result of the differences in transport and also differences in the carbon intensity of energy production and fuel use.9 Assuming BD production would be similar to CDs, carbon equivalent emissions for BDs in 2010 would only fall categorically below downloading for files above 11 GB in the United States (based on lower bound intensity)..
posted by maryr at 2:28 PM on September 4


I wonder if the deliberate network inefficiencies introduced by regional monopolies makes the emissions worse in the USA than in a country like South Korea.
posted by BrotherCaine at 2:36 PM on September 4 [3 favorites]


Do they explain how they came up with 0.3kWh per GB energy use while downloading for "Customer premise equipment (modems/routers)"?

I have fairly slow broadband, at 6 Mbps download. To download one gigabyte at that speed will take me about 22 minutes. For my "modems/routers" to use 0.3kWh in that time, they would need to be running at roughly 1 kilowatt, or one thousand watts. That would mean they're putting out about as much heat as a toaster or hair dryer. They don't. They use more like ten watts all together.

Going the other way: My modem and router use about ten watts, let's say twenty even. I can download about three gigabytes per hour. Each gigabyte then will use 20 watts running for a third of an hour, or about 7 watt hours total, which is 0.007 kWh. And again, their estimate is 0.3 kWh per GB...

If I had faster broadband, which is not at all uncommon, the kWh per GB value would simply decrease linearly with the increase in speed. Maybe they're estimating this for dialup?

It looks like they are off by two orders of magnitude on that estimate, and that makes up a large part of their estimate for the energy use of downloading. Am I missing something? Do they explain that value anywhere? This seems utterly ridiculous, but the paper is peer reviewed, so I really hope I'm missing something.
posted by whatnotever at 2:37 PM on September 4 [25 favorites]


Reading the PDF, I'm having a hard time with the cited kWh/GB figures. They cite 0.3kWh/GB for customer premise equipment, for instance. This is at least an order of magnitude over what I'd expect. Let's assume it takes an hour to download a GB of data to make the math simple. That corresponds to a relatively pokey 2.3Mbps. That'd mean your modem and router have a power usage of 300 watts. That's nuts. A common DOCSIS 3.0 cable modem consumes 8 watts while in use and a common 802.11n wireless router consumes 9.1 watts. Guess I'll go ahead and read the cited study for power usage now to see where they're getting these numbers from.
posted by zsazsa at 2:38 PM on September 4 [6 favorites]


These results seem to be pretty limited by the fact that a PS3 is used as the baseline machine. I'd be curious to see how Steam game distribution fares, as PC's can be MUCH more efficient at power management than most consoles (incl. the PS4 and XBOne). Or even P2P/decentralized game distribution (e.g., torrents), which I imagine is a much more effective means of distributing content because it doesn't require the additional dedicated hardware/network infrastructure of centralized servers.
posted by tybeet at 2:38 PM on September 4


I'd love to read "Malmodin et al. 2014" to see how on earth they're calculating kWh/GB transferred since that's at the heart of the comparison.
posted by Skorgu at 2:38 PM on September 4


I came here to complain about .3 kWh per GB, and on preview other people have done it better. Thanks, other people!
posted by contrarian at 2:40 PM on September 4 [7 favorites]


So howzabout the carbon footprint created by me going to work and earning enough to buy a car so I can then drive to the store and buy an actual Blu-Ray player so I can then purchase this hypothetical game on physical media, huh? I'm probably not even going to like the stupid thing.
posted by Spatch at 2:42 PM on September 4 [5 favorites]


whatnotever: "I have fairly slow broadband, at 6 Mbps download. To download one gigabyte at that speed will take me about 22 minutes. For my "modems/routers" to use 0.3kWh in that time, they would need to be running at roughly 1 kilowatt, or one thousand watts. That would mean they're putting out about as much heat as a toaster or hair dryer. They don't. They use more like ten watts all together.
"

Think! You're not just shoveling those bits across the cable modem and router and discarding them. What's storing that data?
posted by boo_radley at 2:55 PM on September 4 [1 favorite]


I found Malmodin et al. 2014, but it's behind a paywall / intstitutional access wall. I have access through my institution, though...

So they do state that routers and modems use 9-11W. The 0.3 looks like it comes from this: "The CPE’s electricity consumption is nearly the same in active or stand-by state and that is why stand-by energy use should not be overlooked." CPE is "customer premise equipment." If I'm reading that correctly, that means the estimate is for total electricity used over time divided by total downloaded, including all the time the router/modem is just sitting there downloading nothing.

Then the authors say a few sentences later, "Thus, the figures (figure 9) given are an illustration of energy use per amount of data transmitted, but only the figures for data centers, data transmission, and IP core network can be recommended when modeling electricity consumption." The 0.3 kWh/GB figure for CPE, which is not data centers, data transmission, or IP core network is found in... figure 9.

So.
posted by whatnotever at 2:57 PM on September 4 [2 favorites]


boo_radley, they separately account for the energy used by the PS3, which is where the data is stored.
posted by whatnotever at 2:57 PM on September 4


Like 3 people immediately sent me that Malmodin pdf, thanks!
posted by Skorgu at 3:07 PM on September 4


whatnotever, thanks for that.

So then it could be said if 0.3kWh/GB should be counted for the downloads because modems and routers are left on 24/7, it should be counted for the discs as well. Unless the person only buys games on disc because he or she doesn't have internet at home.
posted by zsazsa at 3:07 PM on September 4 [4 favorites]


These results seem to be pretty limited by the fact that a PS3 is used as the baseline machine. I'd be curious to see how Steam game distribution fares, as PC's can be MUCH more efficient at power management than most consoles (incl. the PS4 and XBOne).

Also, you can be using your PC for other things while it's downloading in the background, effectively bringing the power consumption of that part of the system to nearly zero.
posted by anonymisc at 3:17 PM on September 4


Yeah ok Malmodin has these numbers for "Fixed broadband". It's not immediately clear if they're per year or whatever nor what access technology that includes (cable vs DSL for example). I didn't read the whole paper in detail so it might very well be there and I missed it.

Amount of data: 400GB:
User Equipment: "PCs 1.5 desktop/laptop PCs per household, 395 kWh= 1 kWh/GB"
CPE: "CPE setup 1,5 modems/routers per household, 118 kWh = 0.3 kWh/GB"
Access Network: "Average line/sub 31 kWh/line = 0.08 kWh/GB"
Data Transmission and IP core network: "Total access data traffic (ratio in/out about 1:2.5): About 50% of total access traffic was P2P in 2010 (decreases). Data transmission and IP core network in Sweden = 0.08 kWh/GB
Data center data traffic (mainly into IP core, then out to access) is about 40% of total access data traffic Per amount of data traffic in access networks"

There's some more about submarine cables and such but I don't think it's particularly relevant.

Now since this is a large-scale, nation-wide study of usage it (somewhat reasonably) completely ignores the distinction between the fixed and marginal costs of data transfer. A subscriber that consumes 0 bits of data still uses some power. The CPE is almost certainly powered on 24x7, the access network has some slice of capacity that wouldn't need to be purchased, supported or powered for that zero-use customer and even the more central infrastructure has some capacity factor that is being extended by the presence of that customer, even if she transfers no data.

Again, that's pretty reasonable if you're looking at the aggregate usage of power (and hence carbon) of an entire nation, I bet it scales nicely as a predictive factor since on a large enough scale those numbers do seem reasonable.

To use that level of data to make a much more nuanced (and small-scale) point about the marginal difference in power usage on a highly individual scale as paper in the OP does is... unlikely to be particularly accurate. I don't see how this passed peer review.
posted by Skorgu at 3:28 PM on September 4 [7 favorites]


Even on an aggregate level, the total kWh/GB number isn't especially useful for policy decisions-- for example, if everybody got a bit tired of Netflix and the total US bandwidth consumption dropped by 15%, the power usage wouldn't also drop by 15%. Almost every question on a personal or policy level is going to want to use the marginal numbers, because the question is usually of the form "what if we made this proportionally small change relative to the current situation" and not "what if we abandoned this whole internet thing altogether".
posted by Pyry at 3:53 PM on September 4


Yeah it's imperfect at any level but at least at the national level it can answer questions like "if we went from X% to Y% internet penetration what would that do to carbon levels".
posted by Skorgu at 3:57 PM on September 4


But then a number like kWh/household is what you'd want. The kWh/GB is basically only useful in the marginal form.
posted by Pyry at 4:03 PM on September 4


This is why I always drive around for a while after I download something so the carbon footprints cancel out.
posted by ryanrs at 4:04 PM on September 4 [19 favorites]


Hmm: the Acknowledgements and Biographies are interesting in that they reveal that 3 of the authors, including the first and last, are Sony employees:
Kieren Mayers is presently Head of Environment and Technology Compliance at Sony Computer Entertainment Europe.
...
Maria Bauer, at the time of writing, was an environmental officer at Sony DADC, Austria.
...
Amanda Webb is a research engineer at the Center for Environmental Strategy at the University of Surrey, studying the energy use of games consoles working with Sony Computer Entertainment Europe, London, UK.
The article doesn't include any funding source or conflict-of-interest declarations. (The closest it comes is in the Acknowledgements section: "The article is written in [Mayers] private capacity as a researcher at INSEAD. The research does not necessarily reflect the views of Sony, and no official endorsement should be inferred." This is quite different from saying that there was no influence.)

I find it curious that the very surprising results are also ones that happen to favor Sony's DADC production.
 
posted by Westringia F. at 4:09 PM on September 4 [17 favorites]


I have a hard time caring about the carbon foot print of electricity. We use power. We're going to keep using more and more power. It's going to keep getting greener and greener. Physical stuff and driving in non-electric cars, however, clearly costs carbon, and will continue to due so.

For the kWh/GB figure, I wonder if they are assuming a fully-loaded PS3, and I wonder how much a PS3 draws. I'll try to find my kill-a-watt and find out.
posted by Phredward at 4:54 PM on September 4 [1 favorite]


They came up with a lower bound of 2.27kg of CO2 equivalent, which we divide by 0.541 kg/KWh to get 4.2 KWh to download 8 Gigabytes?

Let's say we use 1% of an EC2 server to serve the download for an hour... that is about 0.2 cents, which means less than 60 watt hours at 30cents/kwh if all the charges went to electricity.

The math here doesn't add up.
posted by MikeWarot at 5:07 PM on September 4 [1 favorite]


Here is Kieren Mayers, lead author of the paper, speaking at BBC on "Playstation and the Environment". He starts speaking on this subject around the 9-minute mark.
posted by stbalbach at 5:59 PM on September 4


What about energy to make the boxes and transport the content from a distribution center, physically?? Driving to pick it up ain't the only factor at play.
posted by oceanjesse at 6:06 PM on September 4 [1 favorite]


The only physical place at which I ever buy 10 items at once is the grocery store. In fact, when I buy a game DVD, it almost always involves a trip to the store for the sole purpose of buying that game.
posted by musofire at 6:13 PM on September 4 [1 favorite]


What about energy to make the boxes and transport the content from a distribution center, physically??

True, but what about the energy it takes to dig up the hole to lay the telecom cables, and the energy to produce those cables?

And let's not forget all the energy expelled calling up the telecom/cable companies to deal with service outages, billing issues, and cancellation (with fees).
posted by FJT at 6:39 PM on September 4 [1 favorite]


So the figure for kWh/GB is inclusive of idle electricity utilized by CPE, but the figure for the carbon footprint of buying 10 items (a big assumption) from a store doesn't include the 20 shoppers who drove to the store without buying anything, or who bought unrelated items.

Reading through the data sources and assumptions makes it challenging to take this seriously.
posted by Room 101 at 6:55 PM on September 4 [3 favorites]


The other thing to remember is that the plastic from a Blu-Ray can be recycled, but when you download a game, those zeroes and ones are used up forever.
posted by DirtyOldTown at 6:55 PM on September 4 [25 favorites]


Just for fun, we can look at their estimate for the energy used by the 'Data center/server rooms (“Internet” part)' as well. It is 0 in their lower bound and 1 kWh/GB downloaded in their upper bound. It looks like their upper bound comes from the total energy use of data centers in Sweden in 2009 divided by the total amount of data downloaded in that same time period and place. But of course not all of that energy went toward downloads, as they themselves note. So how much might really have gone to servicing downloads vs other uses of energy?

Well, let's just say we want to host 10GB of data on a CDN and have at least 100 people download it. That's 1,000 GB downloaded, or 1 TB. This handy CDN pricing page compares a few CDNs, one of which will push out 1TB for $50. If we assume all of that price is paying for the electricity, at $0.10 per kWh, and the equipment and infrastructure and maintenance is magically free, then the CDN's datacenters would be using roughtly $50/$0.10 = 500 kWh to serve that much data. That's an upper bound of 0.5 kWh / GB downloaded. And I think it's a really conservative upper bound.

So if we re-do their calculations with an effectively 0 marginal energy usage for the router and modem (because they will be on regardless) and use 0.5 kWh / GB as the upper bound on the data center's energy use, the total lower bound for downloading comes to 0.16 kWh/GB (down from their 0.46) and the total upper bound comes to 0.66 (down from their 1.46). That lower bound, translated to kg CO2, comes in well under their estimate for driving to purchase a physical disc (plus 9 other items). And if you drive to the store just to buy the disc? That ends up well above even the unreasonably high upper bound on the kg CO2 for downloading.

I don't know. I'm sure anyone could pick holes in my research just as well. But I can't help but think that the authors of this are coming to a strong conclusion based on poorly conceived estimates.
posted by whatnotever at 7:07 PM on September 4 [3 favorites]


I would imagine that the carbon footprint varies hugely from region to region within the US (and of course internationally). Electricity is awfully green here in the Pacific NW, with all the dams and whatnot. Not so much in Ohio. If I'm also hitting a server here in the NW (which I probably am -- see above, "cheap electricity"), then I'm feeling OK about that.
posted by gurple at 8:14 PM on September 4 [1 favorite]


Another thing to consider is this seems to focus on console gaming. As a PC gamer who focuses on only 3 or 4 games, if I had to go to the store for large updates, to say nothing of DLC and user-generated mods/content that at times can at times easily exceed 5GB for that particular version of the mod/addon, I'd either not bother in the first place or end up using a lot more fuel.

I would love to see some form of intelligent universal power supply system that could be standardized among manufacturers that could give the user the ability to replace the heat-generating, inefficient power supplies and bricks with one programmable, power conditioned, surge protected, universally grounded box with a battery backup. Heck, make a model that uses those giant 220 volt power connectors you find with washers and dryers in the US - it would be far safer than overtaxing older home electrical wiring, not to mention the money saved by being more efficient and the amount of wasted electricity a smarter system could avoid.
posted by chambers at 8:15 PM on September 4 [1 favorite]


The data they're basing this on is from 2010. I'm pretty sure Moore's law and related factors is continuously changing this in favor of downloading. Bandwidth increases, processing gets faster and uses less power, etc.
posted by Joakim Ziegler at 8:27 PM on September 4 [1 favorite]


chambers: "I would love to see some form of intelligent universal power supply system that could be standardized among manufacturers that could give the user the ability to replace the heat-generating, inefficient power supplies and bricks with one programmable, power conditioned, surge protected, universally grounded box with a battery backup. Heck, make a model that uses those giant 220 volt power connectors you find with washers and dryers in the US - it would be far safer than overtaxing older home electrical wiring, not to mention the money saved by being more efficient and the amount of wasted electricity a smarter system could avoid."

As long as all your stuff is pretty close to the box, this might work. One of the big disadvantages of low-voltage DC power is that the loss in the cables over distances is atrocious, and it gets bad quicker than you think.
posted by Joakim Ziegler at 8:29 PM on September 4 [1 favorite]


Many companies are doing a lot to decrease the carbon footprint of their data centers these days. Such as Apple trying to go 100% renewable according to this article. And Greenpeace rates companies on this too.
posted by Green With You at 8:58 PM on September 4


the lowest carbon footprint is no game at all. Next to that, a deck of cards inherited from your grandparents has an extremely low carbon footprint. And given the last couple weeks in the gaming industry, a ban on video games is looking doubly attractive.
posted by happyroach at 9:01 PM on September 4 [3 favorites]


Does this account for the fact that the routers/switches/infrastructure involved are not solely dedicated to game downloads?
posted by mrbill at 9:12 PM on September 4


Yeah, that 0.3 kW/h for "customer premise equipment" is extremely fishy. Being a huge nerd, I did my own original research:

- Time period: July 2014
- Subject: Myself, a person who lives alone and likes to download and play computer games. On top of this, I was unemployed that month, so I spent a fair amount of time on my PC.
- Place: My apartment, which has gas utilities, and since the weather was mild I didn't use any A/C, so the bulk of my energy usage was my PCs (two desktops, one laptop) and modem/router.

Total energy usage, according to my bill: 162 kWh. Remember, this is for the whole damn apartment, including my lights, stereo, refrigerator, microwave, time I spent playing games but not downloading, etc.

Total data transfer, according to my DD-WRT router: 450 GB (Lots of that was probably games, even! Actually that seems pretty high even for me, some of that is probably LAN bandwidth.)

Total energy usage per GB: 0.36 kW/h. My extremely liberal and unfair calculation is barely above the quoted figure.

Citation fucking needed.
posted by neckro23 at 12:10 AM on September 5 [4 favorites]


Why would you turn on your router and computer (not to mention your lights and other utilities) while you aren't downloading games?
posted by Phssthpok at 12:57 AM on September 5


How did this pass peer review?
Science. It works is cheaper than an ad campaign, bitches!
posted by fullerine at 1:47 AM on September 5


The data they're basing this on is from 2010. I'm pretty sure Moore's law and related factors is continuously changing this in favor of downloading. Bandwidth increases, processing gets faster and uses less power, etc.

Moore's law doesn't really apply since the processing speed and power consumption wouldn't change within the same console over four years unless they bought the newer, slimmer console in that time.

What jumps out at me is that they discount the power difference between hard drive usage (for downloaded games) and Blu Ray drive usage (for physical games) based on the erroneous belief that the system uses one or the other. In fact, most physical games require a certain amount of game data to be installed on the hard drive in order to run smoothly so physical games require the use (and energy consumption) of both the disc drive and the hard drive while downloads utilize only the hard drive during gameplay.
posted by dances with hamsters at 4:55 AM on September 5


a deck of cards inherited from your grandparents has an extremely low carbon footprint.

Well, you do have to kill your grandparents so there is that cost.
posted by srboisvert at 5:43 AM on September 5


"The results of this study are unlikely to change or influence consumer behaviors; the carbon emissions of production and distribution are not known to be factors consumers consider when buying electronic goods or entertainment media."
Probably a good thing. It makes more sense to focus your conservation efforts on the 93% of the lifetime emissions, than to dicker about marginal gains on the 7%.
posted by CBrachyrhynchos at 6:31 AM on September 5


> Place: My apartment, which has gas utilities, and since the weather was mild I didn't use any A/C, so the bulk of my energy usage was my PCs (two desktops, one laptop) and modem/router.

Actually, the bulk of your electricity usage was most likely your refrigerator, making your already high estimate even higher.
posted by Westringia F. at 7:14 AM on September 5 [1 favorite]


> "The results of this study are unlikely to change or influence consumer behaviors; the carbon emissions of production and distribution are not known to be factors consumers consider when buying electronic goods or entertainment media."

In other words, "Dear EU environmental regulators: Please let Sony DADC continue to press disks in Europe. They're 'better' for the environment than downloads are [for very unphysical values of 'better'], and the only way we'll get them to consumers is by making buttloads of them available. Thanks, SonyINSEAD."
posted by Westringia F. at 7:22 AM on September 5 [2 favorites]


> Actually, the bulk of your electricity usage was most likely your refrigerator, making your already high estimate even higher.

Which you said! Sorry for not reading, neckro23.

(You could pretty easily subtract out the fridge contribution if you wanted, though, what with not needing to estimate how frequently you used it since it's on 24/7.)
posted by Westringia F. at 7:30 AM on September 5


So basically, taking the .3 kWh/GB at face value - in other words, that people only have always-on broadband to download games for three or four hours a month - you would have to assume that people only own cars to drive to gamestop. Let's see how THAT comparison plays out.
posted by mellow seas at 7:58 AM on September 5 [2 favorites]


A 2010 comparison of energy/emissions for CDs vs. downloading music online came to a different conclusion.

Granted, they assumed 7 kWh/GB for downloading, but with albums being about 1/100 of the size of an 8 GB game download.
posted by Foosnark at 8:12 AM on September 5


One of the other interesting things that study found was that buying online and having a CD shipped to you has a lower energy and carbon impact than driving to the store.
posted by Foosnark at 8:18 AM on September 5


As long as all your stuff is pretty close to the box, this might work. One of the big disadvantages of low-voltage DC power is that the loss in the cables over distances is atrocious, and it gets bad quicker than you think.

In my case, they are. I suppose I have two areas at my place that have a high "device population" - my entertainment center in my living room, and my office. With two of these power supply systems, DC cables would run no more than 6 feet from a box like I described.

If one were to install such a system in a house, I could see a large central box in a basement or laundry room where 220v would already be there, and running separate heavy AC lines to smaller boxes placed in a few location in the house that would be managed by the main control box where it would be switched to DC there, not unlike an audio snake.

However, the cost and complexity would most likely far exceed the demand, and probably would only be feasible as a component of some 'smart home' kind of system. Getting manufacturers to standardize products with external power supplies and universal connectors would also most likely have to be part of some big environmental law as well. So maybe give it 30 years and it might be possible.
posted by chambers at 10:20 PM on September 5


I measured my PS3's power consumption, and it's roughly 100 watts all the time. The home screen 110, 95 when the screen was off (while opening netflix), and 105 while watching some netflix.
posted by Phredward at 6:00 PM on September 8


« Older "Tulpas are sentient beings imagined into existenc...   |   Geeky women's clothing company... Newer »


You are not currently logged in. Log in or create a new account to post comments.